1

Credit Research

Modelling and Pricing

Hybrid bonds
Julien TURC
Head of Quantitative Research,
Société Générale CIB
 2

Overview

z Common features of hybrid securities

z Pricing hybrid bonds

„ The multiple method
„ SG’s quantitative approach

z Relative value analysis

z Appendices
„ Characteristics of existing issues
„ Sensitivity analysis and stress-testing of the model
„ Technical details on the model
 3

Credit Research

Common features of
hybrid bonds
 4

Subordination, and Coupon deferral

z Subordination of capital
„ Hybrid securities are subordinated to Senior debt in case of a default or
a restructuring
„ Hybrid securities are typically senior against common equity

z Coupon deferral options: each structure is different

„ Optional deferral
z In case no dividends or share buy backs
„ Mandatory deferral
z Usually based on an earnings-related trigger or a solvency ratio (for financial
companies)
 5

Maturity, and Extension

z Hybrid bonds are usually long dated, or perpetual…

z … but can be called by the Issuer:

„ Usually callable after 10 years on every payment date
„ A step-up compensates the investor for the extension risk following the
first call date

z Pricing the call option requires a dynamic framework

 6

Credit Research

Pricing hybrid bonds

 7

A basic approach: the multiple method

z Each hybrid is compared with its CDS which quantifies its credit risk
z New hybrids can be valued by applying the average multiple in the
market
450
Thomson
400

350
Model ASW Spread

300 Lottomatica
Vinci
250

200 Sudzucker
Bayer
150

100 Vattenfall
Michelin
Linde perp Solvay
Siemens Linde 66
50

0 GE Dong Henkel
0 50 100 150 200 250 300 350 400

Market ASW Spread

May 29, 2007: multiple analysis for corporate hybrids

 8

SG’s model: a three-step approach for pricing hybrid

bonds

z Simulate all possible future scenarios on credit spreads and default by

using:
„ The credit curve of the issuer
„ Assumptions on spread volatility

z Determine the company’s decision for each of these scenarios

„ Coupon deferral
„ Extension

z Assign a probability to each of these scenarios and price the product

 9

Simulate all scenarios on credit spreads

z Use the CDS curve to estimate a base case scenario on

spreads
„ CDS curves are extended above the 10y maturity by using the
long-term bond market

z Use a spread volatility to simulate all possible deviations

around this base case scenario on spreads
„ Spread volatility is taken from the index option market
„ It is adjusted for maturity and correlation mismatch
 10

Simulate all scenarios on default

z Default probabilities are implied from the CDS curve

80 15%

Def au lt p r ob ab ilit y
70
CDS Sp d (b p )

60
50 10%

40
30 5%
20
10
0 0%
0 1 2 3 4 5 6 7 8 9 10 0 5 10
CDS Mat u r it y CDS Mat u r it y

z Recovery rate assumptions are needed to compute default

probabilities and default outcomes
„ Senior CDS recovery = 40%
„ Hybrid bond recovery = 0%
 11

Company decision for each scenario

z Coupon risk
„ Mandatory vs. Optional deferral
„ Cumulative vs. Non cumulative deferral

z Extension risk
„ At each possible redemption date, the company decides whether it wants
to call the security depending on the then current price of the security
„ Our model includes a “reputation cost” which prevents the issuer from
calling the bond as soon as it rationally makes sense
 12

Coupon risk: Mandatory deferral

z Simulate a financial ratio

Cash Flow / Sales Reven u e

14%
„ At any interest payment date, coupon is 12%
5 % b ar r ier

automatically deferred if a given financial 10%

ratio stands below threshold 8%

6%

„ We simulate the ratio by a mean-reverting 4%

2%
process fitted to historical data 0%

Financial ratio is correlated with credit

e
95
96
97
98
99
00
01
02
03
04
05
„

spreads
z Compute the probability to hit the deferral
trigger
z Adjust the probability by a risk premium 30%

Mar k et -im p lied d ef au lt

25%

„ Mandatory deferral risk is similar to default

p r ob ab ilit y
20%

risk 15%

10%

„ It is adjusted by the same risk premium as 5%

0%
the default risk premium in the credit 0% 1% = 5% 10% 15% 20%
A- equivalent St at ist ical d ef au lt p r ob ab ilit y
market
 13

Coupon risk: Optional deferral

z Modelling the issuer’s behaviour

„ Optional deferral is linked to the company’s health

z Coupons are deferred when short-term spreads reach a given

threshold
„ 700bp for corporate hybrids

z Historical data on corporates which omitted dividend payments shows

that these are realistic thresholds
„ Although sometimes, dividend has not been restored in spite of a sharp
tightening of spreads…
„ … while in other cases, dividend payments were maintained despite a
credit crisis
 14

Coupon risk: Cumulative vs. Non cumulative

deferral

z Non cumulative: coupons are cancelled

z Cumulative: two possible scenarios:
„ Coupons are deferred and then paid back later
z Losses only come from interests on deferred interests
z Assumption: discarded in our model
„ Coupons are deferred and then the company defaults
z Deferred interests are lost
z Assumption: two years of deferred interests before default
 15

Extension risk

z At each possible redemption date, the company decides whether it wants to

call the security depending on its price
z Reputation cost
„ The company has to face a reputation cost upon extending the security
„ This cost is an additional step-up paid on each coupon date

First call date Redemption date

Fixed payments Floating payments

Euribor Interest
payments to
Floating margin bond holders

Additional loss
Reputation cost suffered quarterly
by issuer
 16

Pricing model

z A Partial Differential Equation (PDE) enables assigning a probability to each

scenario and to compute the Net Present Value of the product in each case
z Numerical method…
„ The hazard rate model: dλt = λt (θ (t ).dt + σ .dWt )
„ Using a binomial tree calibration to extract θ (t ) from the market
„ Using a bootstrapped IR curve to extract r (t ) from the market
z … based on a PDE grid
„ Diffusing only ln(λ ) (IR and ratio risks are considered independent)
„ Finite differences to approximate first and second order spatial derivatives
„ Von Neumann conditions
„ The PDE: X = ln(λ )
∂V ∂V σ 2 σ 2 ∂ 2V
0= + (θ (t ) − ) + − V (r + e X ) + Vdefault e X
∂t ∂X 2 2 ∂X 2
 17

Credit Research

Relative value analysis of

corporate hybrids
 18

Which indicator for corporate hybrid relative value?

z Subordination risk is the major source of risk between hybrids and

senior CDS
„ This is the reason why we use senior recovery rate as a relative value
indicator
„ Hybrid recovery rate is set to 0% while senior recovery rate is adjusted to
fit to market spreads
Sp d Over Ben ch m ar k (b p )

800

600

400

200

0
0% 20% 40% 60% 80%
Sen ior r ecover y r at e
 19

Relative Value analysis

z The pricing model takes into account senior recovery rates for valuing
a hybrid bond
„ Senior recovery rate can be adjusted to market spreads
„ Final model spread is a theoretical spread based on a basket average
senior recovery rate

Pricing
PricingModel
Model

Senior Recovery Rate Hybrid Bond spread

Fit to issuer Implied recovery rate Market spread

Average

Basket average
Fit to basket Model Spread
recovery rate
 20

Pricing corporate hybrids: the multiple approach

z Average multiple was 4.06x in the corporate hybrid market on 12

February 2007
Relative value analysis of corporate hybrids based on multiple vs. CDS

Interpolated CDS Market ASW spd Multiple Fair ASW spd Spd pick-up

Henkel perp 23 144 6.29 93 51

Sudzucker Perp 32 173 5.42 129 43
Solvay perp 23 130 5.60 94 36
Vattenfall Perp 19 111 5.97 75 35
Bayer 2105 33 149 4.52 134 15
Linde 7.375% 66 46 194 4.20 188 6
Dong 3105 26 110 4.26 105 5
Michelin 33 33 132 4.03 133 -1
Linde 6% perp 33 128 3.84 136 -8
Siemens 66 25 82 3.25 102 -20
Vinci Perp 53 190 3.61 214 -24
General Electric 66 16 32 1.95 66 -34
Thomson Perp 88 304 3.46 356 -53
Lottomatica 66 84 250 2.98 341 -91
Tui perp 238 374 1.57 968 -593
Average 51 167 4.06 209 -42

Average multiple
used for the fair spd
 21

Pricing corporate hybrids: SG’s model results

z Implied recovery rate was 33% on May 29, 2007 on the corporate
hybrid market assuming a 200bp reputation cost
Top picks against CDS
Relative value analysis of corporate hybrids
Market ASW Interpolated Implied senior Spread pick-up Last week
Fair ASW spread
spread CDS recovery rate against senior debt spread pick-up

Linde 7.375% 66 172 34 60% 79 93 98

Henkel perp 103 16 74% 24 79 92
Dong 3105 80 16 61% 32 47 48
Solvay perp 94 21 55% 49 45 51
Linde 6% Perp 83 22 46% 53 30 35
Siemens 66 66 21 49% 41 24 38
Michelin 33 98 27 40% 78 20 19
General Electric 66 36 17 29% 40 -4 6
Sudzucker Perp 170 39 31% 175 -5 9
Vattenfall Perp 78 13 19% 94 -16 -7
Bayer 2105 111 24 12% 157 -46 -53
Lottomatica 66 176 84 11% 258 -83 -67
Vinci Perp 162 49 3% 245 -83 -58
Thomson Perp 272 81 0% 425 -154 -137
Tui perp 318 236 0% 872 -555 -547
Average 134 47 33% 175 -40 -32
Source: SG Credit research
Average implied senior
recovery rate Roughly fairly priced Most expensive
issues against CDS
 22

Risk impacts for each issue

z Each hybrid is priced at its fair value (using the average senior recovery
rate) and risk factors are then removed one by one to compute their impact
on the spread
z Extension risk has an average fair value of 59bp, coupon risk of 40bp, and
subordination risk of 22bp
Split of the fair spread of corporate hybrids
Coupon deferral Subordination
Fair ASW spread Extension impact Bullet bond spread
impact impact

Bayer 2105 157 80 36 13 28

Dong 3105 32 2 4 9 19
General Electric 66 40 2 13 7 18
Henkel perp 24 1 -3 7 19
Linde 6% Perp 53 6 12 11 25
Linde 7.375% 66 79 7 16 16 40
Lottomatica 66 258 49 61 43 106
Michelin 33 78 9 24 14 32
Siemens 66 41 5 4 9 23
Solvay perp 49 3 9 11 26
Sudzucker Perp 175 40 66 22 47
Thomson Perp 425 111 188 38 89
Tui perp 872 164 356 102 250
Vattenfall Perp 94 59 6 9 19
Vinci Perp 245 66 91 26 62
Average 175 40 59 22 53
Source: SG Credit research
 23

Credit Research

Relative value analysis of

subordinated financials
 24

Adapting the model to subordinated financials

z Assumptions for subordinated financials

„ Subordinated CDS recovery rate: 20%
„ Recovery rate for LT2: 20%, for UT2: 10%, for T1: 0%

z The model is the same except for dated insurance bonds

„ They have the same subordination as CDS (LT2) therefore they are
not impacted by the CDS recovery rate in the model
„ The reputation cost is used as the adjustment parameter in the
model instead of the CDS recovery rate
 25

Model results for subordinated financials, May 29, 2007

Bank Tier 1 issues with step-up Dated subordinated insurance
y = 0.9008x
50 Hannover Re24-
125
R2 = 0.8741 Zurich 25-15 14

BCP perp-15 40
100
L&G 25-15

Model ASW Spread

Commerz perp-
Model ASW Spread

Munich 23-13
Unicredito perp-
ABN perp-16 16 30 Zurich 23-13
75 15
Allianz 25-15
HBOS perp-16
CDEE perp-15 Prudential 21-11 Std Life 22-12
50 Lodi perp-15 20 Aviva 23-13
ESPSAN perp- Allianz 22-12
14 Axa 20-10
BNP perp-16
Generali 22-12
25 10 Aviva 21-11
Barclays perp-10 RSA 19-09
AIB perp-11

0 0
0 25 50 75 100 125 0 10 20 30 40 50 60
Market ASW Spread Market ASW Spread

Bank Tier 1 issues without step-up Undated subordinated insurance

200
200
180 Hannover Re
180
perp-15
160 160
Model ASW Spread

y = 1.0245x

Model ASW Spread

140
140 R2 = 0.5751
120 Barclays perp-20
Barclays perp-14
120 Std Life perp-15
100 AIB perp-14
Swiss Re perp- Credit Logement
100 80
Old Mutual perp- perp-11
16
60 Dresdner perp-17
Aviva perp-14 Axa 5.77 perp-16 15 CDEE 4.75 perp-
80 AGF perp-15
40 16
60 Allianz perp-17 Generali perp-16 20
Aviva perp-15
40 Allianz perp-14 0
50 60 70 80 90 100 110 120 0 50 100 150 200
Market ASW Spread Market ASW Spread
 26

Credit Research

Conclusion
 27
An exhaustive framework for valuing hybrid
securities

z The model developed by SG is a complete and mathematically

coherent framework for valuing hybrid securities
„ Taking into account all sources of risk affecting hybrid products
„ A quantitative and fundamental approach at the same time

z It can be used to:

„ Find the fair value of options embedded in each structure
„ Analyse the relative value between issues

z Results are updated every week on existing corporate, insurance and

bank subordinated issues
 28

Pavillon d’Armenonville
Thursday July 5 & Friday 6, 2007